Re: [mm][RFC][PATCH 0/11] mm accessor updates.

[Andi Kleen <andi@firstfloor.org>]

On Wed, Dec 16, 2009 at 08:12:18PM +0900, KAMEZAWA Hiroyuki wrote:
> On Wed, 16 Dec 2009 11:49:51 +0100
> Andi Kleen <andi@firstfloor.org> wrote:
> 
> > On Wed, Dec 16, 2009 at 07:31:09PM +0900, KAMEZAWA Hiroyuki wrote:
> > > On Wed, 16 Dec 2009 11:28:06 +0100
> > > Andi Kleen <andi@firstfloor.org> wrote:
> > > 
> > > > > > Also the patches didn't fare too well in testing unfortunately.
> > > > > > 
> > > > > > I suspect we'll rather need multiple locks split per address
> > > > > > space range.
> > > > > 
> > > > > This set doesn't include any changes of the logic. Just replace all mmap_sem.
> > > > > I think this is good start point (for introducing another logic etc..)
> > > > 
> > > > The problem is that for range locking simple wrapping the locks
> > > > in macros is not enough. You need more changes.
> > > > 
> > > maybe. but removing scatterred mmap_sem from codes is the first thing to do.
> > > I think this removing itself will take 3 month or a half year.
> > > (So, I didn't remove mmap_sem and leave it as it is.)
> > 
> > I suspect you would just need to change them again then.
> > 
> Do you have alternative recommendation rather than wrapping all accesses by
> special functions ?

Work out what changes need to be done for ranged mmap locks and do them all
in one pass.

Unfortunately I don't know yet either how exactly these changes look 
like, figuring that out is the hard work part in it.

I suspect it would need changing of the common vma list walk loop
pattern.

-Andi

-- 
ak@linux.intel.com -- Speaking for myself only.